Breast cancer mutation model may predict response to endocrine therapy

SAN ANTONIO — Genomic analyses conducted at baseline and after the initiation of endocrine therapy accurately predicted whether women with ER-positive breast cancer would respond to or progress on treatment, according to study results presented at the San Antonio Breast Cancer Symposium.

“What we need clinically is a better test to predict who will respond to endocrine therapy, and we need to better understand endocrine resistance,” J. Michael Dixon, MBChB, MD, professor of surgery at the University of Edinburgh in Scotland, said during his presentation.

Dixon and colleagues evaluated genomic data from 17 women with ER-positive breast cancer who were post-menopausal. All patients received neoadjuvant letrozole.

Four patients responded well to treatment. The other 13 patients either did not respond to treatment or initially responded but subsequently developed resistance to therapy.

Researchers conducted DNA and mRNA sequencing on 51 tumor samples from the patients. They found that nine patients had the luminal A subtype, seven had luminal B disease, and one patient had HER-2-enriched disease. Three patients who responded had luminal A disease and one had luminal B disease.

The analysis also included genomic data derived from patients at various time points after initiation of endocrine therapy. Overall, patients who progressed on treatment experienced the greatest change in genomic expression. Five patients who progressed were classified as having luminal B disease at baseline, but were then classified has having luminal A disease after progression. Only one patient who progressed was reclassified from luminal A to luminal B disease.

Patients who responded to treatment had high levels of mutations at baseline, but the number of these mutations decreased during treatment, Dixon said. This decrease in mutations did not occur as dramatically in non-responders.

“One of the obvious reasons why you would lose mutations is because you lose cancer cells, but actually we looked at the cancer cells ... and we see that this dramatic fall-off in the number of mutations occurs, but the tumor cellularity remained the same,” Dixon said. “The reason for this is because when you respond to endocrine therapy, the center implodes … so you still get cells out of tumors that are very, very small.”

Researchers identified two genes associated with outcomes that were present at diagnosis — IL6ST and NGFRAP —  and two proliferation genes present at day 14, MCM4 and ASPM. The researchers created a model for treatment response with these mutations that demonstrated 96% accuracy with a 94% positive predictive value and 96% negative predictive value in the training cohort. The model demonstrated 93% accuracy in a validation cohort.

Researchers then applied the model to day-14 data derived from the 17 women in the study cohort. The model was 100% accurate in all women who had static disease and then progressed, who responded to treatment or who did not respond to treatment. The model accurately predicted progression following initial response in five out of six women (83%).

“Baseline analysis is not sufficient to predict response to endocrine therapy or any other therapy,” Dixon said. “We need to analyze the cancer during therapy to see changes. This four-gene model, with two genes at diagnosis and two at day 14, predicts very accurately response to letrozole and appears to have some clinical utility.”

For more information:

Dixon JM. Abstract #S1-05. Presented at: San Antonio Breast Cancer Symposium; Dec. 9-13, 2014, San Antonio.

Disclosure: The researchers report no relevant financial disclosures.